Development of Maximal Dynamic Strength During Concurrent Resistance and Endurance Training in Untrained, Moderately Trained, and Trained Individuals: A Systematic Review and Meta-analysis

A systematic review and meta-analysis was conducted in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-analyses statement guidelines [27]. A search from January 1980 to May 2020 was performed primarily using the PubMed and SPORTDiscus databases. The search strategy used the following combined terms: ‘concurrent strength and endurance training’ OR ‘combined strength and endurance training’ OR ‘simultaneous strength and endurance training’ OR ‘concurrent resistance and endurance training’ OR ‘simultaneous resistance and endurance training’ OR ‘combined resistance and endurance training’ OR ‘cross training strength and endurance’ OR ‘cross training resistance and endurance’. These search terms were used because they were deemed relevant and associated with concurrent training. The title and abstract of the studies identified through the database search were scanned for potential inclusion according to the inclusion criteria. We also manually searched the reference lists of the included papers, whereby eight more studies were included. Studies with insufficient information in the abstract to decide their inclusion or exclusion for final analysis were retrieved for full-text analysis and evaluation. The corresponding authors of the articles were contacted if relevant data were lacking. The search was limited to humans, the English language, and adults.

The following inclusion criteria were applied: (1) randomised and nonrandomised original articles, including healthy normal-weight men and women, 18–40 years of age; (2) interventions compared a group performing lower-body resistance and endurance trainings with a group performing identical resistance training only; (3) participants in the intervention and control groups with an equal baseline training status; (4) resistance training programmes performed in at least two sessions per week, including the same exercises as the main outcome measure of the study, with an intensity > 60% of the 1RM or lighter weight to fatigue; (5) endurance training performed as running or cycling at an intensity > 70% of the maximal heart rate in at least 2 sessions per week; and (6) studies reporting changes in maximal strength in leg press or squat exercise. These exercises were chosen because they are valid and reliable 1RM tests for maximal strength [28, 29], are widely used during lower-limb resistance training, and stimulate the major muscle groups in the lower limbs. The rationale for including both randomised and nonrandomised studies was to maximise data for the final analysis. Age and health restrictions were chosen to enable homogeneous groups of subjects and the potential influence of confounding factors. Interventions with equal resistance training programs were specifically chosen to ensure comparable resistance training stimuli for the intervention and control groups. The rationale for only including studies with a training intensity and duration above certain thresholds was to ensure that the training stimulus would be sufficient to evoke adaptations in both maximal strength and cardiorespiratory fitness [30, 31]. We confirmed that studies that used metrics different from the maximal heart rate met the 70% inclusion criteria by scrutinising the training design characteristics of the endurance training and the prescribed intensities. In studies with obvious error or typographical errors in key outcomes, scientific reasoning and comparisons with other equivalent studies were applied as recommended in the Cochrane Handbook for Systematic Reviews of Interventions, specifically to handle situations where authors are not responding to requests to clarify uncertainties [32]. Scientific reasoning enables sound assumptions about what is missing or is a typographical error. In this report scientific reasoning was used to recalculate the standard deviation (SD) to the standard error of the mean (SEM) or vice versa, where it obviously was a typographical error, or to add information missing in the training design.

To classify the training status, we followed recent recommendations suggesting that training status is a consequence of training history/experience (length of time spent regularly performing an activity or exercising) rather than an objective measure such as 1RM or VO_2max [33, 34]. Objective measures can be added to make a classification more robust but only if the same tests and standardised procedures have been used. As this requirement differed considerably between the studies included in this analysis, especially for measurements of 1RM, we did not include objective measures in the classification process. To distinguish between physical activity and exercise, we used structure, planning, and repetition (regularity) as recommended by the World Health Organization [1].

The following criteria were used to classify participants as untrained, moderately trained, and trained:

In two studies, participants classified as recreationally active by the author were classified as trained, as they were involved in systematic resistance and/or endurance training [35, 36]. To enable a subgroup analysis, the studies were further divided into categories based on whether the resistance and endurance trainings were performed during the same session (< 20 min between exercises) or different sessions (> 2 h between sessions). The < 20-min and > 2-h recovery periods were chosen because none of the studies had periods in between these time points, and they well represent what we considered to be training during the same session as compared with training at different sessions. If a study mixed the same session and different sessions, they were placed into a mixed category and were not included in the subgroup analysis. This applied to two studies [8, 22].

The methodological quality of the included studies was assessed with the PEDro Scale [37]. Only moderate- to high-quality studies (PEDro scores 5–10) were included in the meta-analysis. All the studies were rated according to the PEDro scale by two researchers (HP and HR). The result from the quality assessment can be found in the supplementary information (Appendix S1). Fourteen of the studies [3, 4, 10, 11, 14, 15, 21‐24, 35, 38‐40] provided evidence of moderate quality, and 13 studies [5, 6, 8, 12, 13, 16, 25, 36, 41‐45] presented evidence of high quality.

The mean and SD or confidence interval (CI) was used to present the data. Calculations and analyses were performed with Comprehensive Meta-Analysis (CMA) version 2 (Biostat, Inc., Englewood, NJ, USA). Significance levels were set to p < 0.05. Standard differences in the mean were used to calculate the ES. The threshold for the smallest worthwhile change was set to 0.2. An ES ≥ 0.2 was considered a small effect; ≥ 0.5, a moderate effect; and ≥ 0.8, a large effect [46]. The mean relative change (%) in maximal strength, calculated as the post-training result minus the pre-training result divided by the pre-training value and multiplied by 100, was calculated for both groups (the concurrent resistance and endurance training [CT] group and resistance training only [RT] group). The level of heterogeneity was calculated using the I² statistics. The threshold for heterogeneity was set as follows: an I² value of 25% was considered low; 50%, moderate; and 75%, high [47]. The data were collected from each study and analysed with a random-effects model, and are presented visually in forest plots. The pooled ES with a 95% CI is presented for each category and was compared between the groups. To calculate the ES in CMA, pre- and post-training data for each group (mean, SD, and N) were used. In five of the included studies, the mean and SD were extracted manually from graphs [3, 4, 6, 14, 38]. Funnel plots stratified by training status were used to quantify potential publication bias. As most studies were small-scale in terms of sample size, we also performed an additional sensitivity analysis using Hedges’ g as a complement to standardised difference in means, to evaluate the robustness of our results. A subgroup analysis was performed to examine if the recovery period between the endurance and resistance training sessions could affect the development of maximal strength. To do this, the studies were divided into studies that performed the training in close proximity during the same session (< 20 min apart) and studies that performed the training at different sessions (> 2 h apart).

The database search yielded 1464 potential studies for inclusion (Fig. 1). Twenty-seven studies met the inclusion criteria, and were included in the meta-analysis [3‐6, 8, 10‐16, 21‐25, 35, 36, 38‐45]. A total of 750 participants were included (523 men and 227 women), aged 20–38 years. Seven studies involved untrained individuals [11‐14, 21‐23], 10 studies involved moderately trained individuals [3, 5, 10, 15, 16, 38‐41, 45], and 10 studies involved trained individuals [4, 6, 8, 24, 25, 35, 36, 42‐44]. The corresponding authors of 16 studies were contacted [3‐6, 14‐16, 21, 22, 25, 36, 38‐42] for clarification or missing information via e-mail, of whom five responded with additional information [5, 15, 16, 36, 39]. In two of the included studies, it was not clear in which form the variation (SD and SEM) was reported. Scientific reasoning by comparisons with equivalent studies were then applied to recalculate the variation in the outcomes, as the authors would not respond to requests for clarification [6, 24]. For more detailed information about the participant characteristics, see Table 1.

A summary of training design variables for lower-body strength and endurance training in each study, including recovery between sessions, the sequential order, the frequency and duration of endurance training per session, and the intervention length, is presented in Table 2. The frequency and duration of endurance training per week ranged from 2 to 6 sessions per week and from 12 to 260 min per week. In six of the included studies, information regarding the rest and work durations during intervals was missing. As the authors did not respond with additional information, the endurance duration per session was estimated through scientific reasoning based on the distance that was covered during the endurance training and comparisons with other similar studies [4, 6, 24, 25, 39, 44]. The length of the studies ranged from 6 to 21 weeks. The mean frequency of the endurance training sessions was 2.9 per week for untrained participants, 2.8 for moderately trained participants, and 2.6 for trained participants. The mean duration of the endurance training per session was 37 min for untrained participants, 31 min for moderately trained participants, and 29 min for trained participants. The frequency of strength training ranged from 2 to 5 sessions per week with a mean frequency of 2.9 sessions per week for untrained participants, 2.7 for moderately trained participants, and 2.5 for trained participants. For untrained participants, 4 studies performed endurance training first, and in 3 studies, the intra-session order was not specified. For moderately trained participants, 4 studies performed endurance training first, and 2 performed resistance training first; and in 6 studies, the intra-session order was not specified. For trained participants, 4 studies performed endurance training first, and 5 studies performed resistance training first; and in 3 studies, the intra-session order was not specified.

The endurance exercise type (interval/continuous/mixed) was 0/6/1 for untrained, 9/6/1 for moderately trained, and 11/2/2 for trained participants. Of the studies, 12 performed concurrent resistance and endurance training within the same session (< 20 min between sessions) [4, 5, 13, 14, 16, 21, 23, 25, 36, 40, 43, 45], 13 performed concurrent resistance and endurance trainings during different sessions (> 2 h between sessions) [3, 6, 10‐12, 15, 24, 25, 35, 38, 39, 41, 44], two of the studies mixed performing concurrent resistance and endurance training during the same and different sessions during the training programme [8, 22], and one study did not report whether the trainings were performed in the same or different sessions [42]. With regard to the outcome variables, 15 of the studies measured the maximal dynamic strength with leg press exercise (and two of these also measured squat exercise) [5, 6, 10‐13, 15, 16, 21, 23, 24, 39‐41, 45], and 14 with squat exercise [3, 4, 8, 14, 22, 24, 25, 35, 36, 38, 40, 42‐44].

The strength improvement for the different interventions included in this meta-analysis is presented in Table 3.

The effect of concurrent resistance and endurance training compared with that of resistance training only on the maximal strength for the three categories, untrained, moderately trained, and trained, is shown in Fig. 2. For untrained and moderately trained participants, there were no significant negative effects of concurrent resistance and endurance training as compared with resistance training alone (ES = 0.03, 95% CI  – 0.29 to 0.35; p = 0.87 and ES =  – 0.20, 95% CI  – 0.42 to 0.02; p = 0.08). For trained individuals, there was a small significant negative effect favouring resistance training alone compared with concurrent resistance and endurance training (ES =  – 0.35, 95% CI  – 0.59 to  – 0.11; p < 0.01). There were no indications of heterogeneity in terms of ES as follows: untrained, I² = 0, p = 0.99; moderately trained, I² = 0, p = 1.0; and trained, I² = 1.5, p = 0.43. The relative weight contributions of the included studies were evenly distributed.

A subgroup analysis was performed to compare the ES between studies that performed resistance and endurance trainings within the same session (< 20 min between sessions; Fig. 3) or during different sessions (> 2 h between sessions; Fig. 4). For untrained and moderately trained individuals, there was no significant difference in effect between conducting same session concurrent resistance and endurance training compared with conducting resistance training alone (ES = 0.01; 95% CI  – 0.44 to 0.46; p = 0.98 and ES =  – 0.23, 95% CI  – 0.54 to 0.08, p = 0.14). However, for trained individuals, the results showed a moderately negative effect favouring resistance training alone compared with conducting resistance and endurance training within the same training session (ES =  – 0.66, 95% CI  – 1.08 to  – 0.25, p < 0.01). Low and nonsignificant heterogeneities among the studies were observed for the untrained, moderately trained, and trained individuals (I² = 0, p = 0.94; I² = 0, p = 0.96; and I² = 17.1, p = 0.31, respectively).

Training during different sessions (> 2 h between sessions; Fig. 4) showed no significant difference in effect between concurrent resistance and endurance training and resistance training alone for any of the training status categories (untrained, ES = 0.12, 95% CI  – 0.41 to 0.65, p = 0.65; moderately trained, ES =  – 0.16, 95% CI  – 0.48 to 0.16, p = 0.32; and trained, ES =  – 0.10, 95% CI  – 0.43 to 0.23, p = 0.55). There were no indications of heterogeneity in terms of ES as follows: untrained, I² = 0, p = 0.57; moderately trained, I² = 0, p = 0.95; and trained, I² = 0, p = 0.93.

The present results are in line with those of two recent reviews that proposed that the negative effect of adding endurance to resistance training primarily manifests in trained individuals [17, 33]. A possible explanation for this difference between untrained and trained individuals might be that trained individuals have less potential for adaptations and need more specific training to obtain further performance improvements [48, 49]. In line with this are recent findings showing that block periodisation is superior to a mixed training approach for strength development in athletes [50, 51]. The molecular mechanisms behind this are not well understood, but studies that examined the acute response after resistance exercise showed that trained muscles have a blunted expression of several genes and proteins involved in the anabolic adaptation process as compared with untrained muscles [52‐54]. Moreover, it was recently shown that mTORC1, a major regulator of muscle hypertrophy, is negatively affected by concurrent training in trained but not in untrained individuals [53], and some studies even showed an enhanced molecular response and hypertrophy in untrained and moderately trained participants after concurrent training [23, 52, 55]. Adding endurance training to a resistance training protocol may therefore, under some circumstances, be beneficial for a less trained population.

Another potential explanation for the impaired response observed in the trained category could be a reduced quality of the performed resistance training because of fatigue. Endurance training can lead to acute fatigue and accumulated fatigue over time (over-reaching), leading to reduced intensity or volume of the resistance training performed [56, 57]. Trained individuals have higher aerobic work capacity than untrained individuals and might therefore exhaust themselves more during endurance training, especially during self-regulated high-intensity endurance training. This could potentially lead to greater fatigue and reduced performance during subsequent resistance training sessions. A higher overall workload might also increase the risk of spending more time in a catabolic state, which would be negative for muscle adaptations, especially if endurance and resistance training are performed in close proximity [33]. In addition, studies have shown that trained individuals have a higher potential for voluntary activation of their muscles [58, 59]. This will result in recruitment of a larger proportion of the muscles during a 1RM test, more weight lifted, and a higher relative mechanical tension when training with a load related to this test (for example 80% of 1RM). They might therefore, be able to stress their muscles more during resistance training and consequently need more recovery between the training sessions than less trained individuals [60]. To summarise, the higher overall load/stress from the resistance and endurance training for trained individuals could potentially enhance the interference effect by reducing the quality of the resistance training and blunting the anabolic response.

There are several variables other than the training status that might affect the outcomes of a concurrent training programme. Previous reviews on the topic have identified the following as important: the recovery period between the resistance and endurance training sessions, the sequential order of the endurance and resistance trainings (i.e. endurance before or after resistance), and the frequency, duration, intensity, and modality (i.e. cycling, running, etc.) of the endurance training [61‐65].